Effect of combined consumption of Lactobacillus brevis KB290 and β-carotene on minor diarrhoea-predominant irritable bowel syndrome-like symptoms in healthy subjects: a randomised, double-blind, placebo-controlled, parallel-group trial

Abstract

People with non-pathological diarrhoea-predominant irritable bowel syndrome (IBS-D)-like symptoms are present at a significant level even in healthy populations, but established prophylaxis is lacking. Previously, we have found that co-administration of Lactobacillus brevis KB290 (KB290) and β-carotene (βC) attenuated murine colitis (potential cause of IBS-D-like symptoms) significantly. Here, we investigated the effect of KB290 and βC on minor IBS-D-like symptoms in healthy volunteers. After a 4-week run-in period, subjects received a KB290 + βC or placebo capsule for 12 weeks, followed by a 4-week washout period. The KB290 + βC group showed a significant improvement in intensity of abdominal pain and stool frequency compared with the placebo group. The KB290 + βC group showed a significantly higher serum concentration of anti-inflammatory cytokine, interleukin-10, compared with the placebo group. In conclusion, we demonstrated that consumption of KB290 + βC improves minor IBS-D-like symptoms and inflammatory status in healthy volunteers. (UMIN000018002)

Graphical Abstract

Keywords:

• Lactic acid bacteria
• b-carotene
• gastro-intestinal
• inflammation

View correction statement:

Correction to: Fuke, Aizawa, Suganuma, Takagi, and Naito, Effect of combined consumption of Lactobacillus brevis KB290 and β-carotene on minor diarrhoea-predominant irritable bowel syndrome-like symptoms in healthy subjects: a randomised, double-blind, placebo-controlled, parallel-group trial

Introduction

Irritable bowel syndrome (IBS) is chronic abnormality of the bowel accompanied by pain and/or discomfort in the abdomen. IBS can be subdivided into four types (“constipation”, “diarrhoea”, “alternative” or “unsubtyped”) according to predominant stool consistency. Abdominal symptoms of IBS disturb the daily life (Gralnek et al. 2000) and reduce the work productivity (Paré et al. 2006) of patients. Hence, they not only reduce the quality of life (QoL) of patients but they also affect the economy. In particular, patients with diarrhoea-predominant irritable bowel syndrome (IBS-D) show significantly lower QoL than those with constipation-predominant irritable bowel syndrome (IBS-C) because of abdominal pain-associated diarrhoea and frequent defaecation (Singh et al. 2015).

Causes of IBS-D have not been revealed fully. However, it has been suggested that IBS-D patients have low-grade intestinal inflammation (Spiller et al. 2000; Liu et al. 2014; Vicario et al. 2014). It has also been reported that the severity of intestinal inflammation is related to the severity of IBS symptoms (Di Nardo et al. 2014). Furthermore, it has been demonstrated that colonic inflammation induces IBS-D symptoms in animal models (Qin et al. 2011). Based on these findings, low-grade intestinal inflammation is considered to be a cause of IBS-D.

Probiotics is well known to attenuate intestinal inflammation (Hart et al. 2003), and thus the effects of probiotics on IBS patients have been much investigated. For example, it is reported that Saccharomyces cerevisiae CNCM I-3856 improves abdominal pain/discomfort, and stool consistency in IBS patients (Cayzeele-Decherf et al. 2017). Mixture of three probiotic strains (Lactobacillus paracasei, ssp. paracasei F19, Lactobacillus acidophilus La5 and Bifidobacterium lactis Bb12) also showed favourable effects on IBS symptoms, though there was no statistical significance compared to placebo treatment (Simrén et al. 2010). People distressed by severe IBS symptoms tend to visit hospitals and to take medications (Fujii & Nomura 2008). Conversely, people who have mild, non-pathological, IBS-like abdominal symptoms are present in healthy populations (Krogsgaard et al. 2013) (referred as “IBS-like healthy people” in this study). They are distressed by loss of work productivity and restrictions in daily activity related to abdominal symptoms, but they spend their daily life without any prophylaxis. For IBS-like healthy people, providing functional foods to counteract their abdominal symptoms would be helpful. Some probiotics have effects on abdominal symptoms in IBS-C-like healthy subjects (Guyonnet et al. 2009; Eskesen et al. 2015). Studies on the dietary intervention in IBS-D-like healthy people have not been conducted in-depth.

Lactobacillus brevis KB290 is a lactic acid bacteria isolated from a traditional pickle in Japan called “suguki”. KB290 is recognised as a probiotic because the strain has been verified to improve bowel habits, and to be safe for human consumption (Nobuta et al. 2009). β-Carotene (βC) is a pro-vitamin A contained abundantly in plant-origin foods, and is an important source of vitamin A. Previously, we demonstrated that co-administration of KB290 and βC significantly attenuated murine colitis compared with single administration (Fuke N., unpublished results). Therefore, we hypothesised that combined consumption of KB290 and βC (KB290 + βC) would improve IBS-D-like symptoms via attenuation of colonic inflammation.

In the present study, we investigated the effect of KB290 + βC on mild IBS-D-like symptoms in healthy participants according to the guidance for the clinical trials of drugs for IBS treatment proposed by the US Food and Drug Administration (FDA) (U.S. Department of Health and Human Services, Food and Drug Administration 2012). We also investigated improvement in QoL by KB290 + βC using Short Form (SF)-8 (a method for measuring general QoL) and the Work Productivity and Activity Impairment questionnaire (WPAI; a method for measuring abdominal symptom-related loss of work productivity and impairment in daily life). We also assessed the serum concentration of anti-inflammatory (interleukin (IL)-10) and inflammatory (IL-1β and IL-12) cytokines to ascertain if improvement of IBS-D-like symptoms by KB290 + βC is mediated by attenuation of inflammation.

Study design and methods

Study design

This was a randomised, double-blind, placebo-controlled, parallel-group study carried out in a single clinical centre (Association of Medical Corporation Jikoukai Shiroishi Internal Medicine Clinic, Sapporo) in Japan. KAGOME CO., LTD. (KCL; Tochigi, Japan) prepared the study protocol, and provided study products. All study procedures were undertaken by a clinical research organisation (imeQ inc (IMI) Tokyo, Japan) in the consignment from KCL.

Ethical approval of the study protocol

This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects were approved by the Ethics Committees of KCL and IMI. Written informed consent was obtained from all subjects. The study is registered on the University hospital Medical Information Network Clinical Trials Registry (UMIN-CTR: http://www.umin.ac.jp/ctr/index.htm) as number UMIN000018002 (the direct link to registered study information; https://upload.umin.ac.jp/cgi-open-bin/ctr_e/ctr_view.cgi?recptno=R000020752).

Study products

Because it was suggested that combined administration of KB290 and βC exerted synergistic effect against colonic inflammation (a putative cause of IBS-D like symptoms) in animal model and in vitro assay (Fuke N., unpublished results), the test capsule containing KB290 powder (including lyophilised KB290 and maltitol) and βC powder was produced by KCL. The compounding ratio of KB290 powder in test capsule was determined based on a previous study in which the efficacy of KB290 on IBS symptoms in paediatric patients was demonstrated (Murakami et al. 2012). The βC dose was determined to be ≥3.7 mg/d based on the dose that showed anti-colitis effects in mice (Fuke N., unpublished results). We confirmed that the capsule contained 10 billion CFU (25.7 billion total bacterial cells) of KB290 and 4.0–4.5 mg βC during the present study. Because we previously found that the anti-colitis effect of KB290 was exerted by its cell components (Fuke N., unpublished results), hereafter we unify the description of KB290 contents in test capsule as “25.7 billion bacterial cells”. Starch and calcium stearate were added to each test capsule as excipients. The placebo capsule contained starch and calcium stearate alone. Contents were filled in white capsules so that they were identical in appearance, and were enclosed in a plain aluminium pouch.

Subjects

Eighty healthy Japanese male and female volunteers (20–49 years) were selected from volunteers registered in IMI. Then, they gave written informed consent, detailed their medical history, underwent tests (physiological, biochemical, haematological) and urinalyses. Each of these physiological, biochemical, and haematological tests, and urinalyses was performed by the same operator of Sapporo Clinical Laboratory Inc. (Sapporo, Japan) during the recruitment of subjects. Volunteers who did not meet the exclusion criteria and also met inclusion criteria were enrolled in 1-week screening. Selected volunteers recorded their own abdominal symptoms in a diary (described in the Data collection section below). Then, volunteers who had a lower stool consistency score were excluded. Finally, 44 volunteers were selected as study subjects.

The lone inclusion criterion was subjects suffering from diarrhoea with pain and discomfort in the abdomen, but there were 14 exclusion criteria. The first exclusion criterion was patients likely to have inflammatory bowel disease, gastrointestinal ulcers, pancreatitis, coeliac disease, lactose intolerance, protozoan infection, parasitic infection or other organic gastrointestinal disease. The second criterion was subjects using commercially available drugs or “quasi-drugs” for the prevention and treatment of diarrhoea with abdominal pain and abdominal discomfort on a daily basis. The third criterion was patients currently attending hospital for diarrhoea treatment who had pain and discomfort in the abdomen. The fourth criterion was patients judged by the principal investigator (P. I.) to require medical treatment because of severe diarrhoea with pain and discomfort in the abdomen. The fifth criterion was patients diagnosed with IBS. The sixth criterion was patients who had undergone gastrointestinal surgery (excluding appendectomy). The seventh criterion was women who were breastfeeding or pregnant, or who were planning to become pregnant during the study period. The eighth criterion was patients treated for a systemic disease. The ninth criterion was patients using non-steroidal anti-inflammatory drugs, corticosteroids or antibiotics on a daily basis. The tenth criterion was subjects who were smokers (except those who had stopped smoking for ≥1 year). The 11th criterion was subjects with an abnormal blood test (excluding those in which the range of physiological variation was acceptable according to the PI). The twelfth criterion was subjects who had participated in a clinical trial ≤90 d before the screening test. The penultimate criterion was patients suffering from a mental illness. The final exclusion criterion was subjects who might suffer an allergic reaction to the test food.

Randomisation

Stratified randomisation was performed to balance the placebo and the KB290 + βC group for sex and age. In brief, 44 subjects were divided into women and men, and divided further into higher or lower ages (four categories). Then, subjects were randomly allocated into the KB290 + βC group or placebo group in each category using a random number table. Allocation was operated by a researcher of IMI not involved in taking measurements and analysis.

Blinding

KCL produced study products of identical appearance, and provided them to IMI with a symbolised name. Correspondence of the symbolised name and true name of the product was disclosed from KCL to IMI after completion of data analyses.

Study protocol

After a 4-week run-in period (1–4 W), 22 participants were allocated to the KB290 + βC group, and the other 22 participants were allocated to the placebo group. Then, each participant ingested 1 capsule/d for 12 weeks (5–16 W). Participants were asked to ingest the capsule between the time they arose to the time they ate breakfast. This schedule was followed by a 4-week washout period (17–20 W). After the end of the trial, participants who had experienced adverse events underwent follow-up examination until the disappearance of the events.

Participants recorded their abdominal symptoms in diary every day during the 20-week study period. They also answered the WPAI each week. At the end of run-in, consumption and washout periods, they visited the clinical centre, detailed their symptoms and provided blood and urine for biochemical/haematological tests, measurement of serum levels of cytokines and urinalyses. They also answered the SF-8.

Restrictions on participants during the study were avoidance of (i) an “irregular” lifestyle (lack of sleep, overeating, excessive drinking, poor diet); (ii) consumption of dietary supplements (except for the test capsules); (iii) smoking. Also, participants were asked to retain the quantity/quality of exercise and foods (especially, foods containing vitamin A, βC, lactic acid bacteria or bifidobacteria), and the quantity of sleep.

Data collection

Study outcomes were adopted according to the guidance for the clinical trials of drugs for IBS treatment proposed by the US FDA (U.S. Department of Health and Human Services, Food and Drug Administration 2012). Primary endpoints were stool consistency and intensity of abdominal pain. Secondary endpoints were stool frequency, urgency, abdominal discomfort, distension, incomplete evacuation, straining, passage of gas, borborygmi, “composite score” (sum of stool consistency, abdominal-pain intensity, stool frequency, urgency, abdominal discomfort, distension, incomplete evacuation, straining, passage of gas and borborygmi), SF-8, WPAI and serum concentrations of cytokines.

Stool consistency was assessed at each defaecation using the Bristol Stool Scale: 1 (separate hard lumps, like nuts), 2 (“sausage-shaped” but lumpy), 3 (like a sausage but with cracks on its surface), 4 (like a sausage or snake, smooth and soft), 5 (soft blobs with clear-cut edges), 6 (fluffy pieces with rugged edges, a mushy stool) and 7 (watery, no solid pieces). Intensity of abdominal pain was assessed each day as abdominal pain over the previous 24 h in an 11-point Lickert scale from 0 (no pain) to 10 (worst pain imaginable). Stool frequency was assessed by recording defaecation times each day. Urgency, abdominal discomfort, distension, incomplete evacuation, straining, passage of gas and borborygmi were assessed each day using a five-point Lickert scale from 1 (no symptoms) to 5 (worst symptoms imaginable). The composite score as described above was also calculated. The score of urgency, incomplete evacuation and straining of the day without defaecation was treated as missing data.

General QoL was assessed using SF-8. At the end of each period, participants were asked eight questions about their health status in the past month. Then, a mental-component summary score and a physical-component summary score were calculated as described previously (Fukuhara & Suzukamo 2004).

Abdominal symptom-related loss in work productivity and impairment of daily life was assessed using WPAI-GH v2.2-Japanese with a small modification. Specifically, we changed “health problems” to “abdominal symptoms”; “any physical or emotional problem or symptom” to “diarrhoea, abdominal pain, urgency, abdominal discomfort, distension, incomplete evacuation, straining, passage of gas or borborygmi”. In this questionnaire, subjects answered six questions, which were (i) are you currently employed (working for pay) and, during the past 7 d: (ii) how many hours did you miss from work because of your abdominal symptoms; (iii) how many hours did you miss from work because of any other reason, such as vacation, holidays, or time-off to participate in this study; (iv) how many hours did you work; (v) to what extent did abdominal symptoms affect your productivity while you were working; (vi) how much did abdominal symptoms affect your ability to do your regular daily activities, other than work at a job? Questions (v) and (vi) were assessed using an 11-point Lickert scale from 0 (abdominal symptoms had no effect on my work (or my daily activities) to 10 (abdominal symptoms completely prevented me from working (or doing my daily activities). The questionnaire was recorded each week. Then, overall loss of work productivity and activity impairment were calculated as described previously (Reilly et al. 2004).

Serum concentrations of cytokines were analysed at the end of each period. The concentration of IL-10, IL-1β and IL-12 was assessed using a Human IL-10 ELISA kit (Thermo Fisher Scientific, Waltham, MA), Human IL-1β/IL-1F2 Quantikine^® HS ELISA kit (R&D Systems, Minneapolis, MN) and Human IL-12 p70 Quantikine^® HS ELISA kit (R&D Systems, Minneapolis, MN), respectively. The lower limit of quantification (LLQ; pg/mL) was 0.5, 0.125 and 0.78 for IL-10, IL-1β and IL-12, respectively. If the cytokine concentration was below the LLQ, then the value of the LLQ of the kit was adopted as the datum for the sample.

Safety assessment

The PI assessed the safety of study products based on the results of participant communication, tests (physiological, biochemical, haematological) and urinalyses. The content of the daily diary was also referred to for safety assessment.

Statistical methods

Sample size

Based on a previous study (O’Mahony et al. 2005), we supposed that the difference in the mean score for abdominal pain intensity between treatment and placebo was 0.5, the standard deviation (SD) of the score was 0.5, an α error of a two-tailed test was 0.05 and the power was 80%. Then, we estimated the minimum number of subjects to be 16. Based on a previous study (Fujii & Nomura 2008), we estimated that 4 subjects would dropout from the study because of exacerbation or disappearance of abdominal symptoms. We also estimated that we would need 2 reserve participants. Then, we decided to recruit 44 subjects (22 in each group) for this study.

Statistical analysis

Inter-group comparison was done with the unpaired Student’s t-test at each week for all outcomes. In case of serum concentration of cytokines, two-way repeated measures ANOVA, with time, treatment group, interaction of time × treatment and baseline score as fixed factors for each period, was also done to detect the significance of treatment. Effect size was calculated as Hedges’ g value. Only inter-group comparisons of female:male ratio and work situation were made using Fisher’s exact test. One-way repeated measures ANOVA was used to detect the change over time from baseline (p value was adjusted with the Greenhouse–Geisser correction). Then, post hoc analysis was done with Dunnett’s multiple-comparison test. Inclusion criteria of subjects for statistical analyses showed >80% compliance for prevalence of capsule consumption and entries in the daily diary. Statistical analyses were done as per-protocol analyses using SAS v9.3 (SAS, Cary, NC) with p < .05 (two-tailed).

Primary endpoints

Stool consistency and intensity of abdominal pain were calculated as the mean value for 1 week. Then, changes from baseline were compared between groups. Intra-group comparison with baseline at each week was also undertaken.

Secondary endpoints

Urgency, abdominal discomfort, distension, incomplete evacuation, straining, passage of gas, borborygmi and the composite score were calculated as the mean value for 1 week. Then, changes from baseline were compared between groups. Intra-group comparison with baseline at each week was also carried out. Inter-group comparison of changes in SF-8 score and serum concentrations of cytokines from baseline were undertaken at the end of each period. Inter-group comparison of raw scores of WPAI was also done at the end of each period.

Ad hoc analyses of WPAI

The percentage of participants who had no overall loss in work productivity or no impairment in activity was calculated in each group each week. Then, final data between groups in each period were compared.

Results

Subjects and compliance

This study was carried out July–November 2015. Eighty volunteers were enrolled for screening (Figure 1). Among them, five volunteers refused to be included in the study. Ten volunteers were excluded from the study because of abnormal results in blood tests. Then, screening was carried out with the remaining 65 volunteers. Thereafter, 21 were excluded from the study because of lower stool consistency score. Remaining 44 volunteers were allocated randomly to the KB290 + βC group or the placebo group. During the study period, two subjects in the KB290 + βC group dropped out from the study because they moved from the area. As a result, 20 volunteers in the KB290 + βC group and 22 volunteers in the placebo group completed the study. The prevalence of entries in the daily diary and capsule consumption in participants who completed the study was 94.0% and 88.1%, respectively. Therefore, all 42 subjects met the inclusion criteria for statistical analyses.

Figure 1. Study flowchart. KB290, Lactobacillus brevis KB290; βC, β-carotene.

Baseline characteristics of participants

Baseline characteristics of participants are shown in Table 1. There were no significant differences between the KB290 + βC and placebo group for any baseline characteristic (p ≥ .05). We analysed the correlation between the intensity of abdominal pain and stool consistency or stool frequency in pooled data from all subjects during the baseline week. A significant positive correlation was shown for intensity of abdominal pain and stool consistency (Spearman’s correlation coefficient, ρ = 0.25) (Figure 2(A)), and between the intensity of abdominal pain and stool frequency (ρ = 0.45) (Figure 2(B)).

Figure 2. Correlation between the intensity of abdominal pain and stool consistency or stool frequency at baseline. The score of abdominal pain intensity and stool consistency (A) or abdominal pain intensity and stool frequency (B) of all subjects each day at baseline week was plotted. Spearman’s correlation coefficients were calculated. Black circles denote individual data. Darkness of circles shows overlapping of data. Solid lines indicate collinear approximation. n = 271.

Table 1. Baseline characteristics of participants.

	KB290 + βC (n 20)			Placebo (n 22)
	Mean		SD	Mean		SD	P
Age (years)	34.3		8.2	33.6		9.3	NS
Age range (years)		22–48			21–48
Sex (female/male)		11/9			10/12		NS
Stool consistency	4.72		0.72	4.47		0.85	NS
Intensity of abdominal pain	3.36		1.74	3.33		1.95	NS
Stool frequency (times/d)	1.80		0.73	1.78		0.81	NS
Urgency	2.64		0.75	2.87		0.78	NS
Abdominal discomfort	2.74		0.59	2.9		0.67	NS
Distension	2.54		0.54	2.83		0.76	NS
Incomplete evacuation	2.58		0.66	2.89		0.58	NS
Straining	2.43		0.47	2.47		0.58	NS
Passage of gas	2.90		0.65	2.69		0.63	NS
Borborygmi	2.38		0.69	2.22		0.86	NS
Composite score	28.1		4.07	28.4		5.11	NS
WPAI
Overall loss in work productivity (%)	36.4		30.8	40.5		33.2	NS
Activity impairment (%)	34.2		23.4	43.0		25.6	NS
SF-8
mental component summary	35.9		5.55	37.2		8.17	NS
physical component summary	26.9		6.11	27.0		7.27	NS
IL-10 (pg/mL)	0.64		0.24	0.75		0.29	NS
IL-1β (pg/mL)	0.16		0.06	0.19		0.10	NS
IL-12 (pg/mL)	0.78		0.00	0.78		0.00	NS

Primary endpoints

Hereafter, statistical analyses were undertaken with the data from 20 participants in the KB290 + βC and 22 participants in the placebo group unless stated otherwise. The stool consistency score decreased from baseline in both groups, but there were no significant inter-group differences at any time point (p ≥ .05) (Figure 3(A)). Conversely, the KB290 + βC group showed a significantly lower score of intensity of abdominal pain compared with the placebo group at 13 W and 14 W (Hedges’ g value [95% confidence interval (CI)] = −0.68 [−1.31, −0.05], −0.72 [−1.35, −0.08], respectively) (Figure 3(B)). The same tendency was observed at 11, 15, 16 and 17 W (p < .1). The KB290 + βC group also showed a significant change over time in intensity of abdominal pain from baseline to 20 W (p = .004). Post-hoc Dunnett’s multiple-comparison test revealed that the abdominal pain intensity score was significantly lower at all time points from 5 W to 20 W, except for 6 W, compared with that at baseline in the KB290 + βC group (p < .05).

Figure 3. Effects of KB290 + βC on primary endpoints. Changes in stool consistency score (A) and abdominal pain intensity score (B) are shown. Black and white circles denote mean values for KB290 + βC and placebo groups, respectively. Bars show SD. Baseline is denoted by a dotted line. KB290 + βC group: n = 20, placebo group: n = 22. §p < .1, *p < .05, unpaired Student’s t-test.

Secondary endpoints

Abdominal symptoms

Both groups showed a reduction in stool frequency from that seen at baseline (Figure 4(A)). However, the KB290 + βC group showed a further reduction in stool frequency from 13 W. Conversely, the placebo group showed an increase in stool frequency at that time. Thus, the difference between the groups became noteworthy. The KB290 + βC group showed a lower stool frequency at 17 (p < .1), 18, 19 and 20 W (p < .05) compared with that in the placebo group (Hedges’ g value [95% CI] = −0.62 [−1.25, −0.01], −0.64 [−1.27, −0.02] and −0.70 [−1.33, −0.07] for 18, 19 and 20 W, respectively). A significant change in stool frequency was observed in the KB290 + βC group from that at baseline to 20 W (p = .0005). Post-hoc Dunnett’s multiple-comparison test revealed that stool frequency to be significantly lower at all of time points from 13 W to 20 W compared with that at baseline in the KB290 + βC group (p < .05).

Figure 4. Effects of KB290 + βC on abdominal symptoms in secondary endpoints. Changes in stool frequency (A), urgency (B), abdominal discomfort (C), distension (D), incomplete evacuation (E), straining (F), passage of gas (G), borborygmi (H) and composite score (I) are shown. Black and white circles denote mean values for KB290 + βC and placebo groups, respectively. Bars show SD. Baseline is denoted by a dotted line. KB290 + βC group: n = 20, placebo group: n = 22. §p < .1, *p < .05, unpaired Student’s t-test.

The KB290 + βC group had a sustained distension score during the study period, whereas the placebo group showed a decrease in this score (Figure 4(D)). The KB290 + βC group had a significantly higher distension score compared with that in the placebo group at 8 W and 9 W (p < .05). The same tendency was observed at 17 W (p < .1).

There was no significant inter-group difference for incomplete evacuation (p ≥ .05). However, only the placebo group showed a significant change over time in this score (p = .01) (Figure 4(E)). Post hoc Dunnett’s multiple-comparison test revealed that the incomplete evacuation score was significantly lower at all of time points from 8 W to 19 W compared with that at baseline in the placebo group (p < .05).

There was no significant inter-group difference in the composite score (p ≥ .05). However, only the KB290 + βC group showed a significant change over time in this score (p = .02) (Figure 4(I)). Post hoc Dunnett’s multiple-comparison test revealed the composite score to be significantly lower at 5 W, and from 11 W to 20 W, compared with that at baseline in the KB290 + βC group (p < .05).

There were no significant inter-group or intra-group differences (p ≥ .05) in urgency (Figure 4(B)), abdominal discomfort (Figure 4(C)), straining (Figure 4(F)), passage of gas (Figure 4(G)) or borborygmi score (Figure 4(H)).

General QoL

There were no significant inter-group differences in the mental-component summary or physical-component summary in QoL evaluation using SF-8 (p ≥ .05) (Table 2).

Table 2. Changes in quality of life from baseline.

	Consumption period		Washout period
	Mean	SD	Mean	SD
Mental component summary
βC + KB290	3.1	8.25	0.55	8.85
Placebo	−0.38	9.54	1.15	8.95
Physical component summary
βC + KB290	−3.71	5.83	−2.79	9.19
Placebo	−2.09	5.79	−2.65	5.25

WPAI

In the KB290 + βC group, the number of participants: in full-time employment was 13 (65.0%); employed from the middle of the study was 1 (5.0%); who worked on a random basis was 2 (10.0%); who were unemployed was 4 (20.0%); in the placebo group, the respective numbers were 15 (68.2%), 1 (4.5%), 2 (9.1%) and 4 (18.2%). There were no significant inter-group differences in the work situation of subjects (Fisher's exact test, p = 1). For statistical analyses at each week, the data from subjects who worked during the week were used (KB290 + βC group: n = 13–16; placebo group: n = 14–17). Comparison of the raw scores of WPAI revealed no significant inter-group differences (p ≥ .05) for overall loss in work productivity (Figure 5(A)) or activity impairment (Figure 5(B)).

Figure 5. Effects of KB290 + βC on WPAI. Overall loss in work productivity (A) and activity impairment (B) are shown. Black and white circles denoted mean values for KB290 + βC and placebo groups, respectively. Bars show SD. KB290 + βC group: n = 13–16, placebo group: n = 14–17.

Ad hoc analyses for WPAI

The KB290 + βC group had a higher percentage of participants who showed no overall loss in work productivity (Figure 6(A)) or no impairment in activity (Figure 6(B)) compared with the placebo group throughout consumption and washout periods. Inter-group comparisons of the concluded percentage in run-in, consumption, and washout periods revealed the KB290 + βC group to have a significantly higher percentage of both outcomes in consumption and washout periods compared with that for the placebo group (overall loss in work productivity: Hedges’ g value [95% CI] = 2.91 [2.03, 3.80] and 3.90 [1.08, 6.72] and activity impairment = 1.98 [0.97, 2.98] and 2.69 [0.46, 4.93] for consumption and washout periods, respectively) (Figure 6(C,D)).

Figure 6. Ad hoc analyses for WPAI. Number of subjects who showed no overall loss in work productivity (A) and activity impairment (B) were analysed in each week. Black and white circles denote mean values for KB290 + βC and placebo groups, respectively. Numbers of subjects in each group are the same as those for Figure 5. Data of (A) and (B) were concluded in each period (C and D, respectively). Black and white bars denote mean values for KB290 + βC and placebo groups, respectively. Error bas denote SD. Run-in period: n = 4, consumption period: n = 12, washout period: n = 4. §p < .1, *p < .05, unpaired Student’s t-test.

Serum concentrations of cytokines

The serum concentration of IL-10 decreased from baseline to consumption and washout periods in the placebo group, but it was sustained throughout the study in the KB290 + βC group (Table 3). The KB290 + βC group showed a higher serum concentration of IL-10 compared with the placebo group in consumption and washout periods. Two-way repeated measures ANOVA showed the significant effect of treatment (p = .04). The difference was significant in the washout period (Hedges’ g value [95% CI] = 0.73 [0.10, 1.36]). Conversely, there were no significant inter-group differences in serum concentrations of IL-β or IL-12 (p ≥ .05).

Table 3. Changes in serum concentrations of cytokines from baseline.

	Consumption period		Washout period
	Mean	SD	Mean	SD
IL-10 (pg/mL)
βC + KB290	−0.03	0.14	0.04*	0.16
Placebo	−0.12	0.26	−0.10	0.21
IL-1β (pg/mL)
βC + KB290	−0.03	0.06	−0.04	0.05
Placebo	−0.05	0.10	−0.06	0.09
IL-12 (pg/mL)
βC + KB290	0.01	0.04	0.00	0.00
Placebo	0.00	0.00	0.00	0.00

* p < .05, KB290 + βC group versus placebo group, unpaired Student’s t test.

Adverse events

Unusual events described in the daily diary by participants were constipation, cold, food poisoning, abdominal pain, headache, diarrhoea, metrorrhagia, gastroenteritis, fever, migraine, malaise, stomach ache and itching sensation in the anus. These events were transient and incidental, and were confirmed by the PI not to be correlated with consumption of test capsules. Some subjects showed higher or lower values in laboratory tests compared with respect to standard ranges, but the PI confirmed that these values were within the range of physiological variation, and did not reflect adverse events. One participant showed a moderate level of liver dysfunction at 20 W, but the PI confirmed that this event was not correlated with consumption of test capsules because it occurred only in the washout period, and because the participant reported excessive intake of alcohol. We conducted follow-up tests for this participant, and confirmed recovery from liver dysfunction.

Discussion

We investigated the effect of KB290 + βC on IBS-D-like symptoms in healthy volunteers. Volunteers distressed by IBS-D-like symptoms but who did not require medication were recruited. The stool consistency score of subjects was 4.7 in the KB290 + βC group, and 4.5 in the placebo group at baseline. Stool frequency was 1.8 times/d in both groups. Using the Bristol Stool Scale, a score ≥5 is classified as diarrhoea in Asian populations (Gwee et al. 2010). Indeed, the mean stool consistency score in Japanese IBS-D patients has been reported to be ≥5.7 (Fukudo et al. 2014), ≥5.3 (Fukudo et al. 2016) and ≥5.3 (Fukudo et al. 2016). In addition, the stool frequency of Japanese IBS-D patients has been reported to be 5.1 times/d at disease onset, but 1.5 times/d in a healthy condition (Torii 2008). Taken together, the abdominal symptoms of our participants were extremely mild compared with IBS-D patients, and considered to be within the variations observed in healthy populations.

Abdominal pain and diarrhoea can occur in daily life; these abdominal symptoms do not comprise all the symptoms of IBS. In IBS-D, abdominal pain occurs with worsening of stool consistency and/or increase in stool frequency. In our participants, the intensity of abdominal pain and stool consistency or stool frequency was correlated significantly. Thus, although the abdominal symptoms of our participants were extremely mild, they showed IBS-D-like features.

Among IBS-D symptoms, abdominal pain is reported to have a considerable impact upon QoL (Cain et al. 2006), and thus is recommended to be the primary endpoint for clinical trials in guidance provided by the FDA. In that guidance, it is recommended to assess ‘worst abdominal pain over the past 24 hs’ with an 11-point Lickert scale (0–10). Thus, we evaluated the intensity of abdominal pain according to FDA guidance in the present study. At baseline, the intensity of abdominal pain in participants was 3.4 in the KB290 + βC group and 3.6 in the placebo group, which denoted mild abdominal pain. We demonstrated that consumption of KB290 + βC attenuated the intensity of abdominal pain significantly compared with that in the placebo group. This is first study to assess the intensity of abdominal pain in IBS-D-like healthy people according to FDA guidance, and to demonstrate clearly the efficacy of dietary intervention against this symptom.

Frequent defaecation restricts the activity of IBS-D-like healthy people, and leads to impairment of work productivity and daily life. We demonstrated that consumption of KB290 + βC reduced stool frequency significantly compared with that seen with placebo. Reduction of stool frequency by KB290 + βC was 0.5 times/d from baseline to 20 W. The serotonin receptor antagonist ramosetron is a typical agent given for IBS-D. Ramosetron inhibits peristalsis by blocking serotonin signalling, and clearly improves diarrhoea and frequent defaecation (Min & Rhee 2015). In a clinical trial, maximum reduction of stool frequency by ramosetron was ≈0.5 times/d during the test period. Thus, consumption of KB290 + βC reduced stool frequency at a clinically relevant level in the present study.

The placebo group showed a significantly lower score for distension compared with the KB290 + βC group at some time points due to selective reduction of the distension score in the placebo group. However, the distension score in the placebo group increased significantly from 1 W to baseline (4 W) (one-way repeated ANOVA, p = .04), and decreased thereafter to 20 W at the same level as that seen at 1 W. Thus, the reduction in the distention score in the placebo group was considered to be a false improvement produced by accidental worsening of the symptom at baseline, and did not reflect the inferiority of KB290 + βC against that symptom.

We evaluated QoL using SF-8, but there was no significant difference in the score between the two groups. SF-8 is a tool used to measure general QoL, and reflects the change in QoL regardless of abdominal symptoms. WPAI evaluates the impact of disease-specific symptoms on work productivity and daily activity, and has been used previously for IBS patients (Reilly et al. 2004). However, WPAI has not been applied for IBS-like healthy people. Therefore, the WPAI score evaluated in the present study is valuable to understand the “true” condition of IBS-like healthy people. In our study, overall loss in work productivity and activity impairment were 36.4–40.5% and 34.2–43.0%, respectively. In a study conducted in the USA, Reilly et al. (2004) observed that an overall loss in work productivity was 24.7% and activity impairment was 25.0% in IBS patients with no or mild symptoms. In the same report, an overall loss in work productivity was 37.3% and activity impairment was 45.3% in IBS patients with moderate symptoms. Thus, the loss in work productivity and activity impairment observed in that present study is at the same level with that observed in North American IBS patients with moderate symptoms. As our participants had very mild IBS-like symptoms compared with moderate IBS patients, the results of WPAI scoring might be affected with the differences in occupation, employment status and lifestyle, which are related to the differences in living in different regions and countries. There was no significant inter-group difference in comparisons of overall loss in work productivity and activity impairment. Then, we found that many participants had marked “0%” for each score, suggesting that simply comparing the mean of the raw score was not adequate. In clinical trials for IBS, treatment efficacy is sometimes evaluated as the “number of participants with no symptom” for each symptom. “No symptom” refers to relief from IBS symptoms, and is considered to be a clinically important endpoint. Thus, we assessed the number of participants who described no overall loss in work productivity or no activity impairment in each group. We found that the KB290 + βC group showed a significant improvement of both scores in consumption and washout periods compared with those observed in the placebo group. This was an ad hoc analysis, but results suggested that the improvement in abdominal pain and stool frequency by consumption of KB290 + βC led to an improvement in work productivity and daily activity. Further studies are needed to verify the effect of KB290 + βC on WPAI.

We found that the effect of KB290 + βC on the intensity of abdominal pain and stool frequency was sustained for ≥4 weeks after withdrawal. In contrast, disappearance of this effect after withdrawal has been reported for other treatments. For example, linaclotide improves abdominal pain and stool frequency in IBS-C patients during 12-week consumption, but the scores revert to identical levels as those for placebo after withdrawal (Rao et al. 2012). Linaclotide supplies water to stools by regulation of osmotic pressure via the guanylate cyclase C receptor-mediated release of HCO₃ and Cl⁻ to the intestinal lumen (Layer & Stanghellini 2014). Thus, linaclotide is considered to be symptomatic treatment that “dampens” the stool temporarily, so the effect of linaclotide is not sustained after withdrawal. Conversely, the effect of KB290 + βC was sustained after withdrawal, so KB290 + βC could be prophylaxis for IBS-D-like symptoms rather than symptomatic treatment.

Low-grade colonic inflammation has been considered to be a cause of IBS symptom (Ishihara et al. 2013). In particular, the anti-inflammatory cytokine IL-10 is thought to regulate symptoms because its production is reduced significantly in the colonic mucosa (Macsharry et al. 2008; Gao 2013), stools and serum (Darkoh et al. 2014) of IBS-D patients. Previously, we found that administration of KB290 + βC to mice increased the number of CD11c⁺ macrophages in the lamina propria of the colon (Fuke N., unpublished results). CD11c⁺ macrophages are known to produce high amounts of IL-10 in mice (Rivollier et al. 2012) and humans (González-Domínguez et al. 2015) constitutively. In the present study, we demonstrated that consumption of KB290 + βC increased the serum concentration of IL-10 compared with that seen in the placebo group. Serum concentrations of cytokines are the only indirect markers to assess colonic inflammation. However, our results support the hypothesis that consumption of KB290 + βC improved IBS-D-like symptoms via attenuation of colonic inflammation due to an increase in IL-10 production. Though ethically unsound, biopsies of the colonic mucosae of healthy subjects would enable direct evaluation of colonic inflammation to verify the anti-inflammatory effect of KB290 + βC.

In the present study, the minimum effective dose of KB290 and βC was determined to be 25.7 billion bacterial cells/d and 4.0 mg/d, respectively. Previously, we demonstrated that consumption of KB290 at 59 billion CFU/d for 2 weeks did not elicit adverse events in humans (Nobuta et al. 2009). In general, βC is recognised to not show adverse effects even upon overdose because its absorption is regulated strictly based on the level of vitamin A in blood. However, Druesne-Pecollo et al. (2010) reported that continuous supplementation of βC at 20–30 mg/d could increase the risk of cancer of the lung or stomach in smokers and asbestos workers. Conversely, in the same report, βC supplementation at 6–15 mg/d did not increase the risk of any type of cancer. Bjelakovic et al. (2013) reported that βC supplementation in a dose above recommended dietary allowance (>9.6 mg/d) significantly increased mortality. However, in the same report, βC supplementation in a dose below recommended dietary allowance (≤9.6 mg) did not affect mortality significantly. We also observed no adverse events upon consumption of KB290 + βC capsules for 12 weeks in this study. Taken together, we conclude that consumption of KB290 + βC at the dose used in the present study is safe for humans.

Conclusion

In conclusion, we demonstrated that consumption of KB290 + βC improved the intensity of abdominal pain and stool frequency significantly in an IBS-D-like healthy population, and the effect may have been mediated by inflammation attenuation. Participants in the present study were men and women aged ≤49 years, which is a susceptible age for IBS (Ehlin et al. 2003). However, inflammation-related IBS symptoms have been reported in children (Di Nardo et al. 2014) and the elderly (Darkoh et al. 2014). Thus, we believe that the effect of KB290 + βC on IBS-D-like symptoms could be effective for the general population bothered by these symptoms. Our findings provide important evidence for the development of probiotic products for IBS-D-like healthy individuals.

Acknowledgements

The authors thank all of the volunteers in this study. We also thank Satoshi Sakurai, Dr. Mayumi Yamada, and the other research staff of IMI for driving the study, Dr. Hideaki Hoshino for being the PI, Dr. Masanori Fukao for preparation of study products, Naoko Takase for analyses of study products, Shiroishi Internal Medicine Clinic for the admission and health management of subjects, Sapporo Clinical Laboratory Inc. for laboratory tests, Biomarker Science Co. Ltd. (Kyoto, Japan) for cytokine measurement, Intellim Corp. (Osaka, Japan) for statistical analyses, Dr. Shigenori Suzuki and Takuro Inoue for critical revision of the manuscript, as well as Kunihiko Sato and Shigeo Tamura for critical discussions about the study protocol.

Disclosure statement

N. Fuke, K. Aizawa and H. Suganuma are employees of KCL. N. Fuke and K. Aizawa hold stock in KCL. N. Fuke has a patent pending relating to this work.

Additional information

Funding

The present study was supported by a grant from KCL.